Research paper Testing the MOD-AGE chronologies of lake sediment sequences dated by the 210 Pb method Helena Hercman, Micha1 Ga ˛ siorowski, Jacek Pawlak * Institute of Geological Sciences, Polish Academy of Sciences, Research Center in Warsaw, Twarda 51/55, PL-00818 Warszawa, Poland article info Article history: Received 29 October 2012 Received in revised form 12 December 2013 Accepted 8 January 2014 Available online xxx Keywords: Ageedepth model Randomization LOESS method 210 Pb dating abstract In this study, we tested the MOD-AGE procedure for non-parametric ageedepth model computations for several lake sediment sequences dated by the 210 Pb method. MOD-AGE uses the randomization method (a type of Monte Carlo simulation) for ageedepth model construction and confidence band estimation and the LOESS (locally weighted scatterplot smoothing) method for fitting of the ageedepth function. To validate the constructed ageedepth models, we used (1) sediments of known age and (2) an independent radiometric dating method (radiocarbon dating). All tests confirmed that the MOD-AGE algorithm is a suitable tool for the calculation of ageedepth models for lake sediment sequences. We also tested the MOD-AGE procedure for calculations based on the results of two different dating methods. The tests indicated that an ageedepth model based on both data sets, i.e., probability distributions of 210 Pb dates and radiocarbon dates, is more consistent than the ageedepth models based only on one of the two data sets. The important advantage of MOD-AGE compared to other algorithms is that it models both age and depth value as well as their confidence bands while considering the uncertainties of these values. Ó 2014 Elsevier B.V. All rights reserved. 1. Introduction Paleoenvironmental studies apply many biological, chemical, and physical proxies, usually with high resolutions down to tenths of a millimeter (e.g., XRF scanning). However, it is impossible to date every horizon of collected sediment sequences, and the ability to date a material, the time range of the dating method, and the cost of analysis are the primary restrictive factors. In addition, we also typically require the ages of undated horizons and every point of a sequence (interpolation possibility). Therefore, we must know the continuous relation between age and depth. An accurate chronol- ogy is crucial for many paleoclimatological studies. This is accom- plished using models of ageedepth relationships. An ageedepth model is a specific case of a general problem where two values describing the locations of points in a graph are reported with some uncertainty. In our case, these values are the depths and ages of specific points. Two major problems are associated with the con- struction of an ageedepth model: (1) modeling the age and depth of dated points while reflecting the uncertainty of both modeled values, and (2) choosing a fitting method. The most important features characterizing a reliable agee depth model are the following: (1) modeled values should be re- ported with their uncertainties; (2) the ageedepth function and its confidence bands should be modeled; (3) it should be possible to model the ages of sediments older than the oldest-dated samples, and the uncertainties of this propagation (extrapolation) should be known; (4) the model should not be over-complicated and should represent the reality and remain in agreement with analytical data; and finally, (5) the model should be monotonic. It is known that the age of sediments from a specific depth depends on the deposition rate and post-deposition processes (e.g., resuspension, focusing, sedimentation gaps (hiatuses), dehydra- tion, and mineralization of organic matter). Very often, an agee depth model is simply regarded as a plot of age measurements versus sediment depth (Lotter and Hofmann, 2003; Ga ˛ siorowski and Hercman, 2005) with incorrect assumptions of the linearity of the ageedepth function and no concern for the processes mentioned above. The sedimentation rate is calculated by dividing the sediment thickness by the time of sedimentation. Functions matching the deptheage line (e.g., partial least square) provide a much better approximation of the “true” ratio of sediment thick- ness to time (Fedotov et al., 2004). However, calculations using these methods are not exact. Typically, these models do not take into account the uncertainty of depth and age measurements (e.g., the uncertainty of the activity of measuring a specific isotope). * Corresponding author. E-mail addresses: hhercman@twarda.pan.pl (H. Hercman), mgasior@twarda. pan.pl (M. Ga ˛ siorowski), dzeq@twarda.pan.pl (J. Pawlak). Contents lists available at ScienceDirect Quaternary Geochronology journal homepage: www.elsevier.com/locate/quageo 1871-1014/$ e see front matter Ó 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.quageo.2014.01.001 Quaternary Geochronology xxx (2014) 1e8 Please cite this article in press as: Hercman, H., et al., Testing the MOD-AGE chronologies of lake sediment sequences dated by the 210 Pb method, Quaternary Geochronology (2014), http://dx.doi.org/10.1016/j.quageo.2014.01.001